"A poet once said, 'The whole universe is in a glass of wine.'" -- Richard Feynman, 1963
- Strict Entity–Component–System architecture, No OOP overhead/console-like programming experience/allow unlimited feature module extension.
// the "E"
auto l_testEntity = g_Engine->getEntityManager()->Spawn(false, ObjectLifespan::Scene, "testEntity/");
// the "C"
auto l_testTransformComponent = g_Engine->getComponentManager()->Spawn<TransformComponent>(l_testEntity, false, ObjectLifespan::Scene);
l_testTransformComponent->m_localTransformVector.m_pos = Vec4(42.0f, 1.0f, PI<float>, 1.0f);
// the engine has several "S" to handle the engine-side businesses, while users have the freedom to implement their own systems
class TestSystem : public ISystem {};- Custom container, string and math classes, minimum STL overhead/No 3rd-party math library dependency.
RingBuffer<float> l_testRingBuffer(32);
// All custom containers have a thread-safe version
Array<float, true> l_testThreadSafeArray;
FixedSizeString<64> l_testString;
l_testString = "Hello,World/";
for (size_t i = 0; i < l_testString.size(); i++)
{
l_testRingBuffer.emplace_back((float)i);
}
l_testThreadSafeArray.reserve();
l_testThreadSafeArray.emplace_back(l_testRingBuffer[42]);
auto l_maxPoint = Vec4(l_testThreadSafeArray[0], l_testRingBuffer[1], l_testRingBuffer[16], 1.0f);
auto l_minPoint = Vec4(42.0f, l_testThreadSafeArray[0], -l_testRingBuffer[16], 1.0f);
auto l_testAABB = InnoMath::generateAABB(l_maxPoint, l_minPoint);- Job-graph based parallel task model, fully utilize modern hardware/lock-free in client logic code.
std::function<void()> f_JobA;
f_JobA = []()
{
g_Engine->getLogSystem()->Log(LogLevel::Warning, "I'm worried that C++ would be quite a mess for me.");
};
auto f_JobB = [=](int val)
{
g_Engine->getLogSystem()->Log(LogLevel::Success, "There are always some user-friendly programming languages waiting for you, just search for more than ", val, " seconds you'll find them.");
return true;
};
// Job A will be executed on thread 5 as soon as possible
auto l_testTaskHandleA = g_Engine->getTaskSystem()->Submit("ANonSenseTask", 5, nullptr, f_JobA);
// Job B will be executed with a parameter just after Job A finished
auto l_testTaskHandleB = g_Engine->getTaskSystem()->Submit("NotANonSenseTask", 2, l_testTaskHandleA.m_Task, f_JobB, std::numeric_limits<int>::max());
// Blocking-wait on the caller thread
l_testTaskHandleB.m_Task->Wait();
// Or use with a Producer-Customer model
auto l_testResult = l_testTaskHandleB.m_Future->Get();- Object pool memory model, O(1) allocation/deallocation.
struct POD
{
float m_Float;
int m_Int;
void* m_Ptr;
};
auto l_objectPoolInstance = TObjectPool<POD>::Create(65536);
auto l_PODInstance =l_objectPoolInstance->Spawn();
l_PODInstance->m_Float = 42.0f;
l_objectPoolInstance->Destroy(l_PODInstance);
TObjectPool<POD>::Destruct(l_objectPoolInstance);-
Logic Client-as-a-plugin style, the only coding rule is using the engine's interfaces to write your gameplay code and write whatever you want.
-
The major graphics API support, from OpenGL 4.6 to DirectX 11, from DirectX 12 to Vulkan, and Metal, all supported by one unified interface.
-
Client-Server rendering architecture, supports any kind of user-designed rendering pipeline from the first triangle draw call to the last swap chain presentation.
auto l_renderingServer = g_Engine->getRenderingServer();
// m_RPDC = l_renderingServer->AddRenderPassDataComponent("LightPass/");
l_renderingServer->CommandListBegin(m_RPDC, 0);
l_renderingServer->BindRenderPassDataComponent(m_RPDC);
l_renderingServer->CleanRenderTargets(m_RPDC);
l_renderingServer->BindGPUResource(m_RPDC, ShaderStage::Pixel, m_SDC, 17);
l_renderingServer->DrawIndexedInstanced(m_RPDC, m_quadMesh);
l_renderingServer->CommandListEnd(m_RPDC);
// Execute on separate threads
l_renderingServer->ExecuteCommandList(m_RPDC);
l_renderingServer->WaitForFrame(m_RPDC);-
Physically-based lighting, photometry lighting interface with support of real life light measurements like color temperature, luminous flux and so on.
-
Default rendering client support features like:
- Tiled-deferred rendering pipeline
- Retro Blinn-Phong/Classic Cook-Torrance BRDF (Disney diffuse + Multi-scattering GGX specular) for opaque material
- OIT rendering (w.r.t. transparency and thickness)
- SSAO
- CSM with VSM/PCF filters
- Procedural sky
- Large scale terrain rendering with cascaded tessellation
- Motion Blur
- TAA
- ACES tone mapping
-
Real-time GI, baked PRT for the complex large scale scene or bake-free SVOGI for the limited small scale scene, fully dynamic and lightmap-free.
-
Unified asset management, using popular JSON format for all text data and support easy-to-use binary data I/O.
-
Physics simulation, NVIDIA's PhysX integrated.
-
GUI Editor, Qt-based, easy to extend, easy to modify.
And so on...
All scripts are in /Script folder
Tested OS version: Windows 11 version 22H2
- MSVC 17.53+ (engine-only)
- CMake 3.26+
- Qt Creator 10.0.0+
- MSVC 16.11.25+ (editor-only)
Run following scripts will build Debug and Release configurations in parallel:
SetupWin.ps1
BuildAssimpWin.ps1
BuildPhysXWin.ps1
BuildGLADWin.ps1
BuildEngineWin.ps1
PostBuildWin.ps1- Open
Source\Editor\InnoEditor\InnoEditor.prowith Qt Creator - Change "Projects - Build Settings - General - Build directory" to
..\..\..\Binfor Debug, Profile and Release build configurations - Change "Projects - Run Settings - Run - Working directory" to
..\..\..\Bin - Build the project
Tested OS version: Ubuntu 18.04 LTS
- GCC 8.0 or Clang 7.0 or higher
- CMake 3.10 or higher
- OpenGL library(lGL)
Run following scripts:
echo | SetupLinux.sh
echo | BuildAssimpLinux.sh
echo | BuildGLADLinux.sh
echo | BuildEngineLinux.sh # or BuildLinux-Clang.sh or BuildLinux-CodeBlocks.sh
echo | PostBuildLinux.shTested OS version : macOS 10.13.6, 10.15.4
- CMake 3.10 or higher
- Apple Clang 10.0 or LLVM Clang 8.0 or higher
Run following scripts:
echo | SetupMac.sh
echo | BuildAssimpMac-Xcode.sh
echo | BuildGLADMac-Xcode.sh
echo | BuildEngineMac-Xcode.sh
echo | PostBuildMac.sh- Implement
ILogicClientandIRenderingClientclasses and put the implementation source file insideSource/Client/LogicClientandSource/Client/RenderingClient - Change the CMake variable
INNO_LOGIC_CLIENTandINNO_RENDERING_CLIENTinSource\CMakeLists.txtto your client modules class name - Build engine and launch through
Bin/${BuildConfig}/InnoEditor.exeorBin/${BuildConfig}/InnoMain.exeon Windows, or corresponding executable file on Linux and macOS
- Open the workspace folder in VSCode.
- Set debug launch arguments in
./vscode/launch.json. - Start debug with "Launch" button (default F5)
- Use Atom to load the working copy folder
- Install gcc-compiler package
- Select build/makefile and hit "Compile and Debug" button (default F6)
- (Optional) Change launch arguments in Source/Engine/Platform/LinuxMain/CMakeLists.txt
- Open Build/InnocenceEngine.xcodeproj
- Select "Product" - "Run" (⌘ + R)
Run following script:
BakeScene.ps1 -sceneName [scene file name without extension]-mode [value]
| Value | Notes |
|---|---|
| 0 | engine will handle the window creation and event management, for "slave-client" model like the normal game client |
| 1 | engine requires client providing an external window handle, for "host-client" model like the external editor |
-renderer [value]
| Value | --- | Notes |
|---|---|---|
| 0 | OpenGL | Not available on macOS, currently supported version 4.6 (Deprecated) |
| 1 | DirectX 11 | Only available on Windows, currently supported version 11.4 (Deprecated) |
| 2 | DirectX 12 | Only available on Windows, currently supported version 12.1 |
| 3 | Vulkan | Not available on macOS, currently supported 1.1.92.1 (WIP) |
| 4 | Metal | Only available on macOS, currently supported version 2 (WIP) |
-loglevel [value]
| Value | Notes |
|---|---|
| 0 | print verbose level and all the other higher level log messages |
| 1 | only print success level and higher level log messages |
| 2 | only print warning level and higher level log messages |
| 3 | only print error level log messages |
You need DirectXShaderCompiler to generate compiled shader for runtime
- Run
Scripts/HLSL2DXIL.ps1to generate DXIL file from HLSL shader file for DirextX 12 - Run
Scripts/HLSL2SPIR-V.ps1to generate SPIR-V file from HLSL shader file for Vulkan - Run
Scripts/ParseGLSL.batthenGLSL2SPIR-V.batto generate SPIR-V file from GLSL shader file for OpenGL (Deprecated)
Effective C++: 55 Specific Ways to Improve Your Programs and Designs (3rd Edition)
Inside the C++ Object Model (1st Edition)
Effective Modern C++: 42 Specific Ways to Improve Your Use of C++11 and C++14
API Design for C++ (1st Edition)
Advanced C and C++ Compiling (1st Edition)
Data Structures and Algorithms in C++ (4th Edition)
Game Engine Architecture (1st Edition)
Game Coding Complete (4th Edition)
Real-Time Rendering (4th Edition)
Physically Based Rendering : From Theory to Implementation(2nd Edition)
Computer Graphics with Open GL (4th Edition)
Linear Algebra and Its Applications (3rd Edition)
And more...
Wolfram MathWorld: The Web's Most Extensive Mathematics Resource
Advances in Real-Time Rendering in 3D Graphics and Games
RasterTek - DirectX 10, DirectX 11, and DirectX 12 tutorials
Randy Gaul's Game Programming Blog
And more...
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent